Acoustic mosquito detector to aid campaign against tropical diseases
By John Garth
A new technology is being developed in the war on malaria, dengue and many other tropical diseases — an acoustic sensor that can detect and identify mosquitoes by their whine.
Together with colleagues, Stephen Roberts, RAEng/Man Professor of Machine Learning in the Department of Engineering Science, read an article on how mosquitoes in their mating rituals vary their wingbeats, and therefore the noise they make. They realised that acoustic signatures could be used to not only detect, but to distinguish species of mosquito — potentially a vital tool in disease control, because only certain species carry particular diseases.
Remarkably, the first application is literally an ‘app’ for a mobile phone, cutely called Humbug. But hang on, you might say: what’s the use of developing anything for a device hardly anyone in the tropics possesses? In fact, says Roberts, ‘There are more people with phones in sub-Saharan Africa than have necessarily got access to clean water. It’s a staggering world.’
Oxford is set up very, very well to help bring people together with disparate expertise which fit together to produce something that’s bigger than the sum of the parts
The first goal is simply a detector for mosquitoes, regardless of species. The technical challenge is to detect the whining noise more efficiently than the human ear, and separate it from ambient sounds such as human chatter. Roberts thinks they’ve already achieved this with 90% accuracy.
Already, phones have been set up at locations in Kenya which record automatically when the mosquito whine is detected, then upload the recording to Oxford as soon as a reliable connection is detected. The data collected in the first nine months will be publically released this summer — along with an extensive open-source code base.
The sensor could significantly improve estimates of population size and distribution. These can only currently be inferred from mosquito traps or from infection in humans — and in the case of malaria, with its incubation period of three or four weeks, that is simply too late.
The project has thrived in Oxford’s typically fertile cross-disciplinary brew. ‘Oxford is set up very, very well to help bring people together with disparate expertise which fit together to produce something that’s bigger than the sum of the parts,’ says Roberts.
The data is being calibrated with climatic information, with data about disease incidence from the Malaria Atlas Project, and with vegetation maps pioneered by Kew Gardens.
One goal is to make the technology wearable. Another is to build the necessary mobile phone components into solar lamps, which are magnets for the mosquitoes. Cattle tags have also been mooted. Humbug detectors are due to be more widely tested in the field later this year.
‘Our equipment doesn’t zap mosquitoes, it doesn’t prevent them,’ says Roberts. ‘It merely offers a way of detecting, and potentially giving early warning. We have a poverty of information, and mosquito experts really need this to come up with better solutions.’